I want to make this power supply.
The question is, wouldn't the multiplier capacitor part be better before the LM317?
The question is, wouldn't the multiplier capacitor part be better before the LM317?
Depends on what you want: for low noise, it can probably better be placed where it is right now, for low output impedance or accurate output voltage, where the arrow points to (make sure also to shift C1 and C7 then).
I'm curious what your design goal is. The LM317 has 80 dB of line rejection at 120 Hz (assuming you choose the National Semiconductor version, LM317-N, that's still made by TI), so you have plenty of that. I would probably spend the money and board area on doubling C1 and C7 instead.
If lower noise is your goal, I suggest having a look at the newer LDOs by TI. The TPS7A4700 for example.
If you do want to keep the cap multipliers, I suggest increasing R3 and R4 (and R8, R9) so you get a lower cutoff frequency for the filter in the cap multiplier. With the Darlington configuration of the multiplier transistor you won't have much base current flowing, so those resistors could probably be in the 10-100 kΩ range without issue (other than a larger startup time).
That said, I'm not a huge fan of cap multipliers. That mostly stems from my experience with one when used for ripple rejection in a tube amp. I used a high-voltage MOSFET and needed to protect the gate so it wouldn't blow on startup and power-down. Pete Millett does something similar in his Engineer's Amp. The added protection diodes caused the cap multiplier to stop multiplying every time there was a glitch in the mains voltage, resulting in audible hum on the amp output. That was pretty annoying.
I would include a couple of simulations of what happens if there's a glitch in the mains voltage and/or when the mains voltage is removed. Just to make sure that you don't have excessive negative Vbe or Vce across the multiplier transistors.
Also note that the LM317 and 337 need at least 10 mA of output current to regulate properly. I suggest reducing R1, R6 to 120 Ω.
Tom
If lower noise is your goal, I suggest having a look at the newer LDOs by TI. The TPS7A4700 for example.
If you do want to keep the cap multipliers, I suggest increasing R3 and R4 (and R8, R9) so you get a lower cutoff frequency for the filter in the cap multiplier. With the Darlington configuration of the multiplier transistor you won't have much base current flowing, so those resistors could probably be in the 10-100 kΩ range without issue (other than a larger startup time).
That said, I'm not a huge fan of cap multipliers. That mostly stems from my experience with one when used for ripple rejection in a tube amp. I used a high-voltage MOSFET and needed to protect the gate so it wouldn't blow on startup and power-down. Pete Millett does something similar in his Engineer's Amp. The added protection diodes caused the cap multiplier to stop multiplying every time there was a glitch in the mains voltage, resulting in audible hum on the amp output. That was pretty annoying.
I would include a couple of simulations of what happens if there's a glitch in the mains voltage and/or when the mains voltage is removed. Just to make sure that you don't have excessive negative Vbe or Vce across the multiplier transistors.
Also note that the LM317 and 337 need at least 10 mA of output current to regulate properly. I suggest reducing R1, R6 to 120 Ω.
Tom
Some considerations about cap-multipliers combined with a 317:
https://www.diyaudio.com/community/...grade-any-317-based-v-reg.331491/post-6426362
https://www.diyaudio.com/community/...grade-any-317-based-v-reg.331491/post-6426362